- Introduced `ReachabilityState`, `RuntimeHit`, `ExploitabilitySignal`, `ReachabilitySignal`, `SignalEnvelope`, `SignalType`, `TrustSignal`, and `UnknownSymbolSignal` records to define various signal types and their properties. - Implemented JSON serialization attributes for proper data interchange. - Created project files for the new signal contracts library and corresponding test projects. - Added deterministic test fixtures for micro-interaction testing. - Included cryptographic keys for secure operations with cosign.
8.5 KiB
8.5 KiB
Feed Snapshot Freeze/Staleness Thresholds (SP6)
Status: Draft · Date: 2025-12-04 Scope: Codify feed snapshot governance including freshness budgets, staleness thresholds, freeze procedures, and validation checks.
Objectives
- Define maximum age for vulnerability feed snapshots.
- Establish freeze/thaw procedures for reproducible scans.
- Set staleness detection and alerting thresholds.
- Enable deterministic feed state for replay scenarios.
Freshness Budgets
Feed Types
| Feed Type | Source | Max Age | Staleness Threshold | Critical Threshold |
|---|---|---|---|---|
| NVD | NIST NVD | 24 hours | 48 hours | 7 days |
| OSV | OSV.dev | 12 hours | 24 hours | 3 days |
| GHSA | GitHub | 12 hours | 24 hours | 3 days |
| Alpine | Alpine SecDB | 24 hours | 48 hours | 7 days |
| Debian | Debian Security Tracker | 24 hours | 48 hours | 7 days |
| RHEL | Red Hat OVAL | 24 hours | 72 hours | 7 days |
| Ubuntu | Ubuntu CVE Tracker | 24 hours | 48 hours | 7 days |
Definitions
- Max Age: Maximum time since last successful sync before warnings
- Staleness Threshold: Time after which scans emit staleness warnings
- Critical Threshold: Time after which scans are blocked without override
Snapshot States
┌─────────────┐
│ SYNCING │ ──────────────────────────────────┐
└─────────────┘ │
│ │
▼ ▼
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ FRESH │ ──► │ STALE │ ──► │ CRITICAL │
└─────────────┘ └─────────────┘ └─────────────┘
│ │ │
▼ ▼ ▼
┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ FROZEN │ │ FROZEN │ │ FROZEN │
│ (fresh) │ │ (stale) │ │ (critical) │
└─────────────┘ └─────────────┘ └─────────────┘
State Transitions
| Current State | Trigger | New State | Action |
|---|---|---|---|
| SYNCING | Sync complete | FRESH | Record snapshot hash |
| FRESH | Max age exceeded | STALE | Emit warning |
| STALE | Staleness threshold exceeded | CRITICAL | Block scans |
| CRITICAL | Manual override | CRITICAL (override) | Log override |
| Any | Freeze command | FROZEN | Lock snapshot |
| FROZEN | Thaw command | Previous state | Unlock snapshot |
Snapshot Schema
{
"id": "nvd-2025-12-04T00-00-00Z",
"feed": "nvd",
"source": "https://nvd.nist.gov/feeds/json/cve/1.1/nvdcve-1.1-modified.json.gz",
"syncedAt": "2025-12-04T00:00:00Z",
"hash": {
"blake3": "...",
"sha256": "..."
},
"state": "fresh",
"advisoryCount": 250000,
"lastModified": "2025-12-03T23:45:00Z",
"frozen": false,
"frozenAt": null,
"frozenBy": null,
"thresholds": {
"maxAge": "PT24H",
"staleness": "PT48H",
"critical": "P7D"
}
}
Freeze Procedures
Manual Freeze
# Freeze current snapshot for reproducible scans
stellaops feed freeze --feed nvd --reason "Audit baseline 2025-Q4"
# Freeze with specific snapshot
stellaops feed freeze --feed nvd --snapshot-id nvd-2025-12-04T00-00-00Z
# List frozen snapshots
stellaops feed list --frozen
Automatic Freeze (Replay Mode)
When recording a scan for replay:
- Scanner captures current snapshot IDs for all feeds
- Snapshot hashes recorded in replay manifest
- Replay execution uses frozen snapshots from manifest
{
"replay": {
"feeds": {
"nvd": {
"snapshotId": "nvd-2025-12-04T00-00-00Z",
"hash": "b3:...",
"state": "frozen"
},
"osv": {
"snapshotId": "osv-2025-12-04T00-00-00Z",
"hash": "b3:...",
"state": "frozen"
}
}
}
}
Thaw Procedures
# Thaw a frozen snapshot (resume normal sync)
stellaops feed thaw --feed nvd --snapshot-id nvd-2025-12-04T00-00-00Z
# Force thaw all frozen snapshots
stellaops feed thaw --all --force
Validation Checks
Pre-Scan Validation
# scripts/scanner/validate_feeds.py
def validate_feed_freshness(feed_states: dict) -> ValidationResult:
errors = []
warnings = []
for feed, state in feed_states.items():
age = datetime.utcnow() - state['syncedAt']
thresholds = state['thresholds']
if age > parse_duration(thresholds['critical']):
if not state.get('override'):
errors.append(f"{feed}: Critical staleness ({age})")
elif age > parse_duration(thresholds['staleness']):
warnings.append(f"{feed}: Stale ({age})")
elif age > parse_duration(thresholds['maxAge']):
warnings.append(f"{feed}: Approaching staleness ({age})")
return ValidationResult(
valid=len(errors) == 0,
errors=errors,
warnings=warnings
)
Scan Output Metadata
{
"scan": {
"id": "scan-12345",
"feedState": {
"nvd": {
"snapshotId": "nvd-2025-12-04T00-00-00Z",
"state": "fresh",
"age": "PT12H",
"hash": "b3:..."
},
"osv": {
"snapshotId": "osv-2025-12-04T00-00-00Z",
"state": "stale",
"age": "PT36H",
"hash": "b3:...",
"warning": "Feed approaching staleness threshold"
}
}
}
}
API Endpoints
Feed Status
GET /api/v1/feeds/status
Response:
{
"feeds": {
"nvd": {
"state": "fresh",
"snapshotId": "nvd-2025-12-04T00-00-00Z",
"syncedAt": "2025-12-04T00:00:00Z",
"age": "PT12H",
"frozen": false,
"thresholds": {
"maxAge": "PT24H",
"staleness": "PT48H",
"critical": "P7D"
}
}
}
}
Freeze Feed
POST /api/v1/feeds/{feed}/freeze
Content-Type: application/json
{
"snapshotId": "nvd-2025-12-04T00-00-00Z",
"reason": "Audit baseline"
}
Override Staleness
POST /api/v1/feeds/{feed}/override
Content-Type: application/json
{
"snapshotId": "nvd-2025-12-04T00-00-00Z",
"reason": "Emergency scan required",
"approvedBy": "security-admin@example.com",
"expiresAt": "2025-12-05T00:00:00Z"
}
Alerting
Alert Conditions
| Condition | Severity | Action |
|---|---|---|
| Feed sync failed | Warning | Retry with backoff |
| Feed approaching staleness | Info | Log, notify ops |
| Feed stale | Warning | Notify ops, scan warnings |
| Feed critical | Critical | Block scans, page on-call |
| Feed frozen > 30 days | Warning | Review freeze necessity |
Alert Payload
{
"alert": {
"type": "feed_staleness",
"feed": "nvd",
"severity": "warning",
"snapshotId": "nvd-2025-12-04T00-00-00Z",
"age": "PT50H",
"threshold": "PT48H",
"message": "NVD feed exceeds staleness threshold (50h > 48h)",
"timestamp": "2025-12-06T02:00:00Z"
}
}
Determinism Integration
Replay Requirements
For deterministic replay:
- All feeds MUST be frozen at recorded state
- Snapshot hashes MUST match manifest
- No network fetch during replay
- Staleness validation skipped (uses recorded state)
Validation Script
#!/bin/bash
# scripts/scanner/verify_feed_hashes.sh
MANIFEST="replay-manifest.json"
for feed in $(jq -r '.replay.feeds | keys[]' "${MANIFEST}"); do
expected_hash=$(jq -r ".replay.feeds.${feed}.hash" "${MANIFEST}")
snapshot_id=$(jq -r ".replay.feeds.${feed}.snapshotId" "${MANIFEST}")
actual_hash=$(stellaops feed hash --snapshot-id "${snapshot_id}")
if [[ "${actual_hash}" != "${expected_hash}" ]]; then
echo "FAIL: ${feed} snapshot hash mismatch"
echo " expected: ${expected_hash}"
echo " actual: ${actual_hash}"
exit 1
fi
echo "PASS: ${feed} snapshot ${snapshot_id}"
done
Links
- Sprint:
docs/implplan/SPRINT_0186_0001_0001_record_deterministic_execution.md(SP6) - Spine Versioning:
docs/modules/policy/contracts/spine-versioning-plan.md(SP1) - Replay:
docs/replay/DETERMINISTIC_REPLAY.md